Poverty targeting in public programs: A comparison of some nonparametric tests and their application to Indian microfinance.

Many popular social programs have limited coverage among households at the very bottom of the income and wealth distribution. If a program reaches the poor, but neglects the destitute, the (pre-program) income distribution of participants and non-participants will cross. We are interested in the statis-tical methods that can be used to test for this particular pattern of program participation. Our numerical simulations suggest that recently developed tests for distribution crossing are powerful even when the two distributions under study are fairly similar and they can be usefully combined with more stan-dard quantile tests to characterize program participation among the very poor. We apply this approach to data on household expenditures and membership of micro-credit groups in India and find that participation among the poorest households in the study area was lower than that of slightly richer households.